Abstract
Automation trust was explored in a human-in-the-loop Air Traffic Control simulation. Controllers were asked to detect closely approaching aircraft pairs, assisted by an automated tool whose reliability varied according to whether the scenario contained Free Flight (aircraft deviating from flight plans at will). Training on the causes of variable reliability, given to half the controllers, enabled them to appropriately trust the automation less (assessed subjectively) during Free Flight. The trained group also had higher Hit and False Alarm rates than the other group, and were more likely overall to unquestioningly accept the tool's judgments; this behavior sometimes aided task performance, but the training did not affect overall performance. A new System Confidence rating scale showed promise as a supplement to established Trust and Self Confidence ratings, as did a scale enabling participant reports of the degree of automation information use in making each decision. Implications for decision support design and training are discussed.
Get full access to this article
View all access options for this article.